The nervous system poses unique challenges to gene expression analysis because of its extreme cellular heterogeneity and complex distributions of messenger RNAs within individual cells. Indeed the number of neural phenotypes is largely unknown. We propose to develop new methods for analyzing neural phenotypes by using cDNA from neurons that are fluorescently labeled in vivo as targets for hybridization to microarrays. These methods will require the coordinate development of new hardware and software tools. Toward that end, a collaborative program among molecular biologists and computational biologists is proposed that will: 1) develop new hardware for microarrays by comparing the performance of Gel pad microarrays, provided as part of collaborative studies with the Motorola Corporation, to alternative methodologies. Gel pad microarrays represent a new technology that has several theoretical advantages relative to glass microarrays or arrays of oligonucleotides 2) develop methods for the preparation of cDNA from single neurons or at least single neuronal types. cDNA will be prepared from neurons that have been marked in vivo in transgenic mice in which the expression of fluorescent proteins has been using modified BACs. Different "color" fluorescent proteins are first being to neurons expressing NPY (Cyan Fluorescent protein) or POMC (Topaz Fluorescent protein) 3) develop new for tracking, management, querying and retrieval of generated expression data and identification of expression data points within the collected images. In addition, new algorithms for clustering cell by analyzing the level of expression of a neuronal RNAs to those in a pool of CY3 labeled RNA from ten mouse organs. These algorithms will also incorporate data from a set of 100 unchanging RNAs identified and independently quantitated (Specific aim 3). It is widely believed that the use of cDNA microarrays will have an enormous impact on biological research. This impact is likely to be especially great in efforts to study the nervous system where even the total number of cell types is not entirely clear. The methods proposed in this are designed to advance this technology to establish the phenotype of neural cells.